MNRAS 000,1–11 (2020) Preprint 18 November 2020 Compiled using MNRAS LATEX style file v3.0

LOFAR Detection of a Low-Power Radio Halo in the Cluster Abell 990

N. D. Hoang ,1★ T. W. Shimwell,2,3 E. Osinga,3 A. Bonafede,4 M. Brüggen,1 A. Botteon3, G. Brunetti,5 R. Cassano5, V. Cuciti 1 A. Drabent6, C. Jones7, H. J. A. Röttgering,3 and R. J. van Weeren3

1Hamburger Sternwarte, University of Hamburg, Gojenbergsweg 112, 21029 Hamburg, Germany 2Netherlands Institute for Radio Astronomy (ASTRON), P.O. Box 2, 7990 AA Dwingeloo, The Netherlands 3Leiden Observatory, Leiden University, PO Box 9513, NL-2300 RA Leiden, the Netherlands 4Department of physics and astronomy, Bologna University - Via Zamboni, 33, 40126 Bologna, Italy 5IRA INAF, via P. Gobetti 101 40129 Bologna, Italy 6Thüringer Landessternwarte, Sternwarte 5, 07778 Tautenburg, Germany 7Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA

Accepted 2020 November 13. Received 2020 October 15; in original form 2020 May 18

ABSTRACT

Radio halos are extended (∼ Mpc), steep-spectrum sources found in the central region of dynamically disturbed clusters of . Only a handful of radio halos have been reported 14 to reside in galaxy clusters with a mass 푀500 . 5 × 10 푀 . In this paper we present a LOFAR 144 MHz detection of a radio halo in the Abell 990 with a mass of 14 푀500 = (4.9 ± 0.3) × 10 푀 . The halo has a projected size of ∼700 kpc and a flux density of 20.2 ± 2.2 mJy or a radio power of 1.2 ± 0.1 × 1024 W Hz−1 at the cluster (푧 = 0.144) which makes it one of the two halos with the lowest radio power detected to date. Our analysis of the emission from the cluster with Chandra archival data using dynamical indicators shows that the cluster is not undergoing a major merger but is a slightly disturbed system with a mean 44 −1 temperature of 5 keV. The low X-ray luminosity of 퐿푋 = (3.66 ± 0.08) × 10 ergs s in the 0.1–2.4 keV band implies that the cluster is one of the least luminous systems known to host a radio halo. Our detection of the radio halo in Abell 990 opens the possibility of detecting many more halos in poorly-explored less-massive clusters with low-frequency telescopes such as LOFAR, MWA (Phase II) and uGMRT. Key words: galaxies: clusters: individual (Abell 990) – galaxies: clusters: intra-cluster medium – large-scale structure of Universe – radiation mechanisms: non-thermal – diffuse radiation

1 INTRODUCTION 1.4 GHz), steep-spectrum1 (훼 . −1) sources that pervade large regions of the cluster. They are apparently unpolarised down to a Diffuse (∼Mpc) synchrotron radio emission (i.e. halos) in the centre arXiv:2011.08189v1 [astro-ph.HE] 16 Nov 2020 few percent above ∼1 GHz. In turbulent re-acceleration model (e.g. of galaxy clusters indicates the presence of relativistic particles and Brunetti et al. 2001; Petrosian 2001; Brunetti & Lazarian 2007), large-scale magnetic fields in the intra-cluster medium (ICM). It is radio halos are generated by merger-induced turbulence that ac- well known that the acceleration of particles to relativistic energy celerates cosmic ray (CR) particles and/or amplifies the magnetic and the amplification of magnetic fields at these scales are associated fields in the ICM. In addition, hadronic models that attempt to ex- with dynamical events in the host cluster of galaxies. However, the plain the generation of relativistic particles by the inelastic collisions physical mechanisms governing the particle acceleration and the between relativistic protons and thermal ions in the ICM seems to magnetic field amplification have not been fully understood (e.g., play a lesser role in the formation of radio halos (Ackermann et al. Brunetti & Jones 2014; van Weeren et al. 2019). 2010; Brunetti & Jones 2014; Ackermann et al. 2016; Brunetti et al. Radio halos are extended (∼Mpc), faint (∼μJy arcsec−2 at 2017), but could be a more significant process in mini-halos that

★ E-mail: [email protected] 1 the convention 푆 ∝ 휈 훼 is used in this paper

© 2020 The Authors 2 N. D. Hoang et al. have smaller sizes (. 500 kpc; Pfrommer & Enßlin 2004; Zandanel covering 27 percent of the northern sky will be released in the et al. 2014; Brunetti & Jones 2014). upcoming Data Release 2. A990 was observed with LOFAR, as Observational evidence shows that the generation of radio ha- part of LoTSS, during Cycle 4 and is covered by three pointings: los depends on the dynamical state of the host clusters and their P154+50, P156+47 and P158+50. Details of the observations are mass (e.g. Wen & Han 2013; Cuciti et al. 2015). Radio halos are given in Table1. more likely to be observed in clusters that are massive and highly The LOFAR data was calibrated in two steps to correct disturbed. for the direction-independent and direction-dependent effects us- The power of radio halos is known to roughly correlate with ing Prefactor2 (van Weeren et al. 2016; Williams et al. 2016; their cluster mass and X-ray luminosities (e.g. Cassano et al. 2013; de Gasperin et al. 2019) and ddf − pipeline3(Tasse 2014a,b; Bîrzan et al. 2019). The scatters in the correlations reflect the evo- Smirnov & Tasse 2015; Tasse et al. 2018; Shimwell et al. 2019), lution of halos and dynamical state of clusters (e.g. Cuciti et al. respectively. We briefly outline the procedure below. For a full de- 2015). Spectral properties of the halos are also a contributing factor scription of the procedure, we refer to Shimwell et al.(2019). to the scatters. According to turbulent re-acceleration models these In the direction-independent calibration, the absolute ampli- properties reflect the efficiency of particle acceleration and are con- tude is corrected according to the Scaife & Heald(2012) flux nected with cluster mass and dynamics. Steep spectrum halos are scale. The calibration solutions are derived from the observations of generally expected in less massive systems and at higher redshift, 3C 196 which model has a total flux density of 83.1 Jy at 150 MHz. and are thought to be less powerful sitting below the radio power The initial XX-YY phase and clock offsets for each station are de- - cluster mass correlation (e.g. Cassano et al. 2006, 2010b, 2013; rived from the amplitude calibrator and are transferred to the target Brunetti et al. 2008). data. The target data are flagged and corrected for ionospheric Fara- Our understanding of the formation of radio halos are mainly day rotation4. During the data processing, the radio frequency in- based on studies of galaxy clusters that are more massive than terference (RFI) is removed with AOFlagger (Offringa et al. 2012) 14 5 × 10 푀 . There have been only a handful of radio halos de- when needed. The data are then phase calibrated against a wide- 14 tected in clusters that are less massive than 5 × 10 푀 (Abell field sky model that is extracted from the GMRT 150 MHz All-sky 545, Giovannini et al. 1999; Bacchi et al. 2003; Abell 141 Duch- Radio Survey (TGSS-ADR1) catalogue (Intema et al. 2017). esne et al. 2017; Abell 2061, Rudnick & Lemmerman 2009; Abell The direction dependent calibration step mainly aims to correct 2811, Duchesne et al. 2017; Abell 3562, Venturi et al. 2000, 2003; for the directional phase distortions caused by the ionosphere and Giacintucci et al. 2005; PSZ1 G018.75+23.57, Bernardi et al. 2016; the errors in the primary beam model of the LOFAR High Band An- RXC J1825.3+3026, Botteon et al. 2019; Abell 2146, Hlavacek- tennas (HBA). The direction dependent corrections are solved for Larrondo et al. 2018; Hoang et al. 2019). This is due to the sen- each antenna in kMS5 (Tasse 2014a,b; Smirnov & Tasse 2015) im- sitivity limitation of previous radio observations that are unable to plemented in the ddf − pipeline. The flux densities of the LOFAR detect faint radio halos in low-mass clusters. As a consequence, detected sources are corrected with the bootstrapping technique us- radio halos in the regime of low-mass clusters are largely unex- ing the VLSSr and WENSS catalogues (Hardcastle et al. 2016). plored. Steep-spectrum nature of radio halos makes low-frequency Prior to the bootstrapping, the WENSS catalogue is scaled by a fac- telescopes such as LOw Frequency ARray (LOFAR; Haarlem et al. tor of 0.9 to be consistent with the flux density scale in LoTSS (i.e. 2013) ideal instruments for studying radio halos in the low-mass Scaife & Heald 2012). To improve the image quality, the pipeline regime. uses a “lucky imaging” technique that generates additional weights In this paper, we present our search for extended radio emission to the visibilities basing on the quality of the calibration solutions from the galaxy cluster Abell 990 (hereafter A990). With a mass of (Bonnassieux et al. 2018). For imaging, DDFacet imager (Tasse 14 푀500 = (4.9 ± 0.3) × 10 푀 (Planck Collaboration et al. 2016), et al. 2018) is used to deconvolve the calibrated data on each facet. A990 is an excellent target to search for cluster-scale radio emission To improve the quality of the final images in the cluster region, in a mass range that is below the typical one where radio halos are the data are processed through “extraction” and self-calibration found. We make use of the LOFAR Two-metre Sky Survey (LoTSS; steps. In the extraction, all sources outside of the cluster region Shimwell et al. 2017, 2019) 120 − 168 MHz data. The Karl G. are subtracted because they are not of the interest of this work. The Jansky Very Large Array (VLA) 1 − 2 GHz data and the archival sources outside of a 400×400 box centred at the cluster are subtracted NRAO VLA Sky Survey (NVSS) 1.4 GHz data are used to constraint using the directional calibration solutions from the ddf − pipeline the spectral properties the extended radio sources. In addition, we run. After the subtraction, data from all pointings are phase-shifted also use the archival X-ray Chandra data (Obs ID: 15114) to study to the cluster position, averaged in time and frequency, and com- dynamical state of the cluster. pressed with Dysco (Offringa 2016) to reduce data size. The data are We assume a flat ΛCDM cosmology with Ω푀 = 0.3, ΩΛ = corrected for the primary beam attenuation in the direction of the −1 −1 0 0.7, and 퐻0 = 70 km s Mpc . With the adopted cosmology, 1 cluster by multiplying with a factor that is inversely proportional to corresponds to a physical size of 151.6 kpc at the cluster redshift, the station beam response. The requirement for the beam correction i.e. 푧 = 0.144. The luminosity distance to the cluster is 퐷퐿 = here is that the region of interest (i.e. the angular size of the cluster) 682.1 Mpc. is small enough for the station beam response to be approximately uniform. In addition, the total flux density in the extraction region should be above ∼0.3 Jy for the calibration to be converged. Details

2 OBSERVATIONS AND DATA REDUCTION 2.1 LOFAR data 2 https://www.astron.nl/citt/prefactor LoTSS is an on-going 120–168 MHz survey of the entire northern 3 https://github.com/mhardcastle/ddf-pipeline hemisphere (Shimwell et al. 2017, 2019). As of 4th March 2020, 4 https://github.com/lofar-astron/RMextract 1522 of 3168 pointings have been observed. The calibrated data 5 https://github.com/saopicc/killMS

MNRAS 000,1–11 (2020) LOFAR Detection of a Radio Halo in Abell 990 3

Table 1. Observation details

Telescope LOFAR 144 MHz푎 VLA 1.5 GHz

Project/Pointing LC4_034/P154+50, P156+47, P158+50 15A-270 Configurations HBA_DUAL_INNER B Observing dates Aug. 08, Jul. 21, Jun. 08, 2015 Feb. 22, 2015 Obs. IDs L345594, L351840, L345592 - Calibrators 3C 196 3C 196, 3C 286 Frequency (MHz) 120 − 168 1000 − 2000 Bandwidth (MHz) 48 1000 On-source time (hr) 24 0.6 Integration time (s) 1 3 Frequency resolution (kHz) 12.2 1000 Correlations XX, XY, YX, YY RR, RL, LR, LL Number of stations/antennas 59, 60, 59 26

Notes: 푎 for a detail description of the LoTSS data, see Shimwell et al. 2017, 2019.

Table 2. Imaging parameters. observed for four scans of approximately ten minutes each, with a few hours in between the scans. The calibration is processed in the 8 푎 standard fashion (flux, phase, amplitude, band-pass etc.) with CASA Data 푢푣-range Robust 휃FWHM 휎 (k휆)(uv − taper)(00 × 00, 푃 퐴푏)(μJy beam−1) using 3C 147 and 3C 286 as calibrators. The flux density scale was set according to the Perley & Butler(2017) flux scale. The RFI was LOFAR 0.15 − 65 −0.25 16.6 × 8.9 80 automatically removed by the CASA ‘tfcrop’ and ‘flagdata’ tasks. To 00 ◦ (5 )(84 ) eliminate the remaining RFI after the CASA flaggers, the aoflagger 0.15 − 65 0.25 60.4 × 46.7 180 (Offringa et al. 2012) was employed. Additionally, outlier visibilities 00 − ◦ (10 )( 73 ) were looked for in the amplitude-푢푣 distance plane by inspecting VLA 0.52 − 74.6 0.25 10.9 × 8.6 38 the cross-hand (RL, LR) correlations. Since most radio sources (1000)(68◦) are generally not polarised, the cross-hand correlations are a good Notes: 푎: Briggs weighting of the visibility. 푏: position angle. additional indicator of RFI. In this plane, visibilities more than 6휎 away from the mean were flagged. of the “extraction” and self-calibration steps are described in van Finally, to remove residual amplitude and phase errors and Weeren et al.(2020). increase the quality of the image, we performed self-calibration The combined data sets are processed with multiple self- of the target field. Three rounds of phase calibration and three calibration iterations including ‘tecandphase’ and gain calibration rounds of amplitude and phase calibration with the decrease of the rounds. After each iteration, the calibration solutions are smoothed solution interval each round were done. Amplitude solutions with with LoSoTo6 (de Gasperin et al. 2019) to minimize the effect of a signal-to-noise ratio smaller than three were flagged, as well as noisy solutions. The final intensity images of the cluster at different outlier (> 3휎 away from mean) amplitude solutions. The remaining resolution are created using WSClean7 (Offringa et al. 2014) with RFI is manually flagged before the calibrated data is imaged with WSClean the imaging parameters in Table2. To enhance extended sources, (Offringa et al. 2014). The final images are corrected for we subtract compact sources from the 푢푣 data and make images the attenuation of the primary beam by dividing the image pixel CASA of the cluster with more weightings on the short 푢푣 baselines. The values by the VLA primary beam model created by . models for compact sources used in the subtraction are obtained from re-imaging the cluster using only data with 푢푣 distance longer than 1.0 푘휆, which do not contain extended emission larger than 2.3 Chandra data 0 4.2 (i.e. 637 kpc at the cluster redshift). Chandra observations of A990 were performed for 10 ks on June To examine the flux scale in the LOFAR images, we compare 03, 2013 with the AXAF CCD Imaging Spectrometer (ObsID: the flux densities of nearby compact sources with those in the TGSS- 15114). The data were calibrated using the ClusterPyXT auto- ADR1 150 MHz survey (Intema et al. 2017). The selected sources matic pipeline9 (Alden et al. 2019) that calls pre-built routines in have a flux densities above 20휎 in both LOFAR and TGSS-ADR1 CIAO10(Fruscione et al. 2006) and Sherpa11(Freeman et al. 2001). images. We found that the flux scale in our LOFAR image is about The pipeline is able to generate X-ray surface brightness, spec- 4 percent higher than the scale in the TGSS-ADR1 image. We adapt tral temperature, pressure, particle density, and entropy maps from in our analysis a flux scale uncertainty of 10 percent. Chandra observations with minimal inputs from users. The pipeline retrieves the data and corresponding backgrounds from the Chandra data archive. As our interest is X-ray extended emission from the 2.2 VLA data cluster, point sources are manually removed from the data. High The VLA L band (1 − 2 GHz) data was taken on February 22, 2015 while the telescope was in the B-array configuration. The target was 8 https://casa.nrao.edu 9 https://github.com/bcalden/ClusterPyXT 6 https://github.com/revoltek/losoto 10 https://cxc.harvard.edu/ciao 7 https://gitlab.com/aroffringa/wsclean 11 https://cxc.cfa.harvard.edu/sherpa/

MNRAS 000,1–11 (2020) 4 N. D. Hoang et al. energy events are identified and filtered out by the pipeline. The In this paper, we adopt the flux density measured with the ≥ 3휎 gas temperature and pressure are obtained by spectral fitting of the pixels for the radio halo in A990 (i.e. 푆144 MHz = 20.2 ± 2.2 mJy). events in the regions of interest. ClusterPyXT uses Astronomi- The radio halo is undetected in the NVSS and VLA high- cal Plasma Emission Code (APEC) model for optically thin colli- frequency images in Figure3. The non-detection could be due to sionally ionized host plasma and a photoelectric absorption model the sensitivity of the NVSS and VLA data. To examine this pos- (i.e. X−Spec PHABS) that includes redshift, , temperature, sibility, we compute the expected surface brightness at 1.4 GHz normalization and hydrogen column density along the line of sight and 1.5 GHz, and compare it with the measurements in the NVSS (Balucinska-Church & McCammon 1992; Smith et al. 2001). We and VLA images. Using the flux density measurement from the use a metallicity of 0.3 푍 (e.g. Werner et al. 2013) and a hydrogen LOFAR data, we estimate the flux densities for the halo to be 20 −2 column density of 1.31 × 10 cm (Ben Bekhti et al. 2016) for 푆1.4 GHz = 2.1 mJy and 푆1.5 GHz = 2.0 mJy. Here we assume a the ICM. For details of the reduction procedure, we refer to Alden flat spectrum, 훼 = −1.0, for the halo. This is a conservative ap- et al.(2019). proach as radio halos are generally found to have steeper spectra (훼¯ ≈ −1.3; Feretti et al. 2012). The true flux densities of the halo at 1.4 GHz and 1.5 GHz could be lower than these estimates (e.g. with 훼 = −1.3, 푆1.4 GHz = 1.1 mJy and 푆1.5 GHz = 1.0 mJy). To calcu- 3 RESULTS late the expected surface brightness of the halo, we assume that the halo emission is uniformly distributed over the source area. Hence, 3.1 Radio emission −2 the surface brightness is 퐼1.4 GHz = 푆1.4 GHz/퐴 = 0.035 μJy arcsec −2 2 In Figure1 we present the LOFAR 144 MHz radio emission contour and 퐼1.5 GHz = 0.033 μJy arcsec , where 퐴 = 61, 075 arcsec is maps of A990 at high and low resolution (see Table2 for the image the area of the halo (i.e. covered by the ≥ 3휎 in the LOFAR map). properties). The high-resolution contours show the detection of However, the sensitivity for the NVSS and VLA images is 휎 = −2 −2 radio emission from galaxies which are labelled in Figure2. The 0.20 μJy arcsec (Condon et al. 1998) and 휎 = 0.36 μJy arcsec radio emission from the (BCG; at 훼 = (see Table2), respectively. These are not sensitive enough to detect h m s d m s the halo in A990. However, the radio emission at 144 MHz is not 10 23 39.9 , 훿 = +49 08 38.4 and redshift 푧푠 푝 = 0.14216 ± 0.00005; Alam et al. 2015) is also seen in the high-resolution image. constant over the halo (Figure3). The surface brightness emission Due to the low surface brightness of the cluster-scale emission, it is at the peak region is ∼1.7 times the mean pixel value in the halo only seen in the low-resolution contours. In the subsections below, region. In case that the emission at 1.4 GHz and 1.5 GHz follows the we present measurements for the detected sources. similar structure at 144 MHz, the surface brightness in some regions of the halo is expected to be higher, i.e. up to 0.06 μJy arcsec−2 and 0.05 μJy arcsec−2, respectively. The expected surface brightness is still below the detection limit of the NVSS and VLA observations. 3.1.1 Extended radio source We also note that the NVSS and VLA observations are not ideal for Figure 1 shows that the extended radio emission is detected in detecting faint, extended emission at the scales of the cluster. The the cluster centre, and its morphology is mainly extended in the NVSS data obtained from snapshot observations is poorly sampled NW−SE direction. The minimum and maximum projected sizes in the 푢푣 space. The VLA 1.5 GHz observations in B configuration within the 3휎 contour are 515 kpc and 960 Mpc, respectively. is also lacking of short baselines.To estimate the halo spectrum, The projected size of the extended source is approximated as future deep observations at high frequencies with the VLA (e.g. C, 퐷 = √︁515 kpc × 960 kpc ≈ 700 kpc. The location and size of the D configuration) and uGMRT or at low frequencies with LOFAR extended source that mostly overlays the X-ray extended emission Low Band Antennas (10–80 MHz) will be necessary. suggest that it is a radio halo. The integrated flux density for the radio halo measured in 3.1.2 Compact radio sources the LOFAR low-resolution map is 푆144 MHz = 20.2 ± 2.2 mJy. Here we select only pixels that are detected with ≥ 3휎 in the A number of compact radio sources are detected with our LOFAR elliptical region in Figure2. We did not integrate the emission in observations in Figure1. In Table3 we give the flux densities and the region of source G and H as this is likely the residuals of the spectra between 144 MHz and 1.5 GHz for the sources within a imperfect subtraction of these sources. The error takes into account field of view of ∼100 centred on the cluster. The spectral indices the image noise and the flux scale uncertainty of 10 percent which of the compact sources range from −0.50 to −1.84. The are added in quadrature. The corresponding 144 MHz power for the of sources B, D, K, and P range from 0.1326 to 0.15073 implying 24 −1 radio halo at 푧 = 0.144 is 푃144 MHz = (1.2 ± 0.1) × 10 W Hz that they are at similar redshifts of the cluster members (푧¯ = 0.144). (푘−corrected, assuming a spectral index of 훼 = −1.3, i.e. the mean The overlaid radio-SDSS optical images of the compact sources indices for a number of known halos, Feretti et al. 2012). To the are shown in Figure A1. Sources A , E, G, H, I, L, and N do not best of our knowledge, A990 hosts one of the lowest radio power have clear SDSS optical counterparts. Sources A and B consist of halos at low (. 300 MHz) frequencies known to date (together with multiple point sources. Sources G and H are likely two lobes of a RXC J1825.3+3026 studied by Botteon et al. 2019). If the halo FR-I type . has a steep spectrum, 훼 . −1.3, the radio power at 1.4 GHz is 푃 (6.1 ± 0.7) × 1022 W Hz−1 which is still being one of 1.4 GHz . 3.2 X-ray emission the lowest powers for a radio halo at 1.4 GHz (Botteon et al. 2019; Bîrzan et al. 2019). When all pixels above 2휎 are included, the The Chandra 0.5 − 2.0 keV image of the cluster is shown in Fig- values increase by about 12 percent to 푆144 MHz = 22.7 ± 2.4 mJy ure4. The X-ray emission from the cluster is relatively concentrated 24 −1 and 푃144 MHz = (1.3 ± 0.1) × 10 W Hz . A similar difference in in the central region. Its peak emission is closed to the location of the flux density measurements within 2휎 and 3휎 contours was also the BCG. In the outskirts, the X-ray emission has a lower sur- seen in PSZ2 G099.86+58.45 (i.e. 15 percent; Cassano et al. 2019). face brightness and is elongated in the NE−SW direction, which

MNRAS 000,1–11 (2020) LOFAR Detection of a Radio Halo in Abell 990 5

500 kpc + 49:12:00 + + + + 10:00 + + BCG + + 08:00 Declination + + + + + 06:00 +

04:00

10:24:00 23:45 30 15 Right Ascension

Figure 1. LOFAR 144 MHz high- (16.600 × 8.900, cyan) and low- (60.400 × 46.700, yellow) resolution contours on the SDSS optical (i, r, and g band) image. The contours are drawn at levels of ±[1, 2, 4, 8, 16, 32] × 3휎, where 휎 = 80 μJy beam−1 and 휎 = 180 μJy beam−1 for the high- and low-resolution contours, respectively. Negative contours are shown by the dotted lines. Compact sources including the BCG marked with the plus (+) are subtracted in the low-resolution contours. The synthesis beams are shown in the bottom left corner. Source labels are given in Figure2. is perpendicular to the major axis of the LOFAR 144 MHz radio data to derive morphological indicators and to construct a cluster emission. In the NE peripheral region of the cluster, X-ray emission temperature map. is found, but radio emission is not detected. Reversibly, extended radio emission is detected in the west side of the cluster without 3.2.1 Dynamical status X-ray emission seen. The total X-ray luminosity between energy range 0.1 keV and 2.4 keV within an elliptical region in Figure Morphological properties of X-ray emission are known to correlate 44 −1 4 is 퐿X = (3.66 ± 0.08) × 10 ergs s , implying that A990 is with dynamical state of the host clusters of galaxies (e.g. Cassano one of the lowest X-ray luminosity clusters ever found to host a et al. 2010b; Bonafede et al. 2017). The X-ray emission from A990 radio halo (Bîrzan et al. 2019). In the energy range 0.5–2.0 eV, the in Figure4 is mostly concentrated in the central region and is 44 −1 X-ray luminosity is 퐿X = (2.45 ± 0.06) × 10 ergs s . The ellip- slightly elongated in the NE−SW direction which seems to indicate tical region is chosen to roughly cover the X-ray 3휎 contours (i.e. a slightly disturbed cluster but without major merging activities. We 휎 = 1.5 × 10−9 counts cm−2s−1 ). At the measured X-ray luminos- estimate a set of parameters to classify the merging status. These ity the cluster is about 8 times underluminous in the radio power include (i) the concentration of the X-ray emission 푐 (Santos et al. compared with the 퐿X − 푃1.4 GHz correlation (e.g. Cassano et al. 2008), (ii) the centroid shifts between the X-ray emission peak and 2013). the centroid of the X-ray emission within aperture sizes (e.g. Poole

et al. 2006), and (iii) the power ratios 푃푚/푃0 (푚 = 3; Buote & Tsai To study dynamical status of the cluster, we use the X-ray 1995). These are briefly described below.

MNRAS 000,1–11 (2020) 6 N. D. Hoang et al.

Table 3. Flux densities and spectra for radio compact sources.

Source 푆144 MHz [mJy] 푆1.4 GHz [mJy] 훼 푧 A 1.77 ± 0.19 0.04 ± 0.04 −1.66 ± 0.41 – B 1.79 ± 0.20 –– 0.14216 ± 0.00005푎 C 75.10 ± 7.51 4.27 ± 0.22 −1.26 ± 0.05 0.6283 ± 0.0356푎 D 3.07 ± 0.32 0.34 ± 0.04 −0.97 ± 0.07 0.1326 ± 0.027푎 E 255.57 ± 25.46 34.74 ± 1.74 −0.88 ± 0.05 – F 3.08 ± 0.32 0.44 ± 0.04 −0.86 ± 0.06 0.5383 ± 0.0303푎 G 2.33 ± 0.25 0.71 ± 0.05 −0.53 ± 0.06 – H 1.26 ± 0.15 0.18 ± 0.04 −0.87 ± 0.11 – I 1.31 ± 0.15 0.01 ± 0.04 −2.18 ± 1.81 – J 0.44 ± 0.09 –– 0.6717 ± 0.0655푎 K 0.59 ± 0.10 –– 0.1326 ± 0.0364푎 L 1.24 ± 0.15 0.24 ± 0.04 −0.73 ± 0.09 – M 1.52 ± 0.17 0.02 ± 0.04 −1.88 ± 0.79 0.1219 ± 0.0206푎 N 1.08 ± 0.13 0.14 ± 0.04 −0.92 ± 0.14 – O 0.24 ± 0.08 0.21 ± 0.04 −0.06 ± 0.17 0.095 ± 0.0481푎 P 0.93 ± 0.12 –– 0.15073 ± 0.00012푏

References: 푎: Alam et al.(2015); 푏: Rines et al.(2013).

constrain the dynamical state of galaxy clusters (e.g. Böhringer et al. 500 kpc 2010; Cassano et al. 2010b; Bonafede et al. 2017; Savini et al. 2019). Clusters that are dynamically disturbed and host radio halos have 49:12:00 H 푃 G a low value of 푐, high values of 푤 and 3/푃0 (e.g. Cassano et al. I 2010b; Cuciti et al. 2015), except some cases in Bonafede et al. J (2014, 2015a). 10:00 F Following the calculations in Cassano et al.(2010b); Bonafede et al.(2017), we estimate the morphological parameters for A990 to B K −7 be 푐 = 0.18, 푤 = 0.025, and 푃3/푃0 = 2.17 × 10 . The X-ray emis- sion peak used in the calculation is shown in Figure4. The centroid 08:00 A Declination shift is calculated for circular apertures within 푅 = 500 kpc. The E ap D C first aperture starts from 푟푖 = 0.05푅ap and the next ones increase in steps of 0.05푅ap. The power ratio 푃3/푃0 is also calculated out to a 06:00 O L radius of 푅ap. We add our results to the morphological parameter P N diagrams in Figure6. The morphological parameter diagrams in Figure6 show that M A990 is located in the morphologically disturbed quadrant sepa- 04:00 rated by the medians of 푐, 푤, and 푃3/푃0 calculated from a sample of 32 galaxy clusters with X-ray luminosity 퐿 ≥ 5 × 1044 erg s−1 10:24:00 23:45 30 15 푋 Right Ascension and redshift 0.2 ≤ 푧 ≤ 0.32 (Cassano et al. 2010b). When adding more clusters from Bonafede et al.(2015b) and from double relics Bonafede et al.(2017), the medians are slightly shifted (see Figure Figure 2. LOFAR 144 MHz high-resolution image of A990. The high-(gray) 6) and the points for A990 are closed to the boundaries for dis- and low-(blue) resolution contours are the same as those in Figure1. The turbed and relaxed clusters. This is in line with the idea that radio ellipse is the region where the flux density of the halo is integrated. Compact halos are generated in disturbed galaxy clusters that supply energy sources are labelled. via turbulence for the particle re-acceleration and magnetic field amplification/generation.

i. Concentration of X-ray emission is defined as 푐 = ( ) 푆X 푟1<100 kpc /푆X (푟2<500 kpc), where 푆X (푟푖) is the total X-ray sur- 3.2.2 Temperature map face brightness within radius 푟푖. We use ClusterPyXT package to perform spectral fitting of the ii. Centroid shift 푤 = 푆퐷 [Δ푖 (푟푖 )]/푅ap, where Δ푖 is the projected separation between the X-ray emission peak and the centroid of the X-ray data and to produce temperature map in Figure7. The tem- th perature map shows that the ICM gas has a mean temperature of i aperture of radius 푟푖. th about 5 keV which is a moderate value for the known clusters (e.g. iii. Power ratio 푃푚/푃0 presents the fraction of the m multipole moment to the total power gravitational potential. The lowest power Frank et al. 2013; Zhu et al. 2016). Across the cluster, the distri- ratio moment with 푚 = 3 gives a measure of the cluster substructures bution of the ICM gas temperature is patchy. The temperature is (e.g. Böhringer et al. 2010). For detailed formula, we refer to Buote slightly higher in the NW region (i.e. ∼5.4 keV) than that in the SE & Tsai(1995). region of the cluster (i.e. ∼5.1 keV). In the core region, the mean temperature is about 5.2 keV around the X-ray emission peak and Several studies have used these morphological parameters to slightly increases to 5.4 keV at the distance of 120 kpc. However,

MNRAS 000,1–11 (2020) LOFAR Detection of a Radio Halo in Abell 990 7

0.0 0.5 1.0 1.5 2.0 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 0.0 0.2 0.4 0.6 0.8 1.0 (mJy/beam) (mJy/beam) (mJy/beam) LOFAR 144 MHz NVSS 1.4 GHz VLA 1.5 GHz

500 kpc 500 kpc 500 kpc

Figure 3. (From left to right) LOFAR 144 MHz, NVSS 1.4 GHz, and VLA 1.5 GHz image of A990. The NVSS and VLA (grey) contours are ±[1, 3, 9, 27, 81]× 3휎, where 휎 = 450 μJy beam−1 and 38 μJy beam−1 for the NVSS and VLA images, respectively. The field of view of the images and the LOFAR low-resolution (yellow) contours without compact sources are the same as those in Figure1.

26 10 A990 ( = 1.3) 1 2 3 4 5 6 7 counts cm 2 s 1 1e 8 25

) 10 1 z 49:12:00 500 kpc H W ( 24

z 10 H G 4 . 1 P

1023 10:00

+ 1044 1045 L (erg s 1) 08:00 500 Declination 1026 A990: ( = 1.3)

25

) 10

06:00 1 z H W (

z 24 H 10 G 4 . 1 P 10:24:00 23:48 36 24 Right Ascension 1023

Figure 4. Chandra 0.5 − 2.0 keV image (bin=2) of the cluster overlaid 1015 with the LOFAR low-resolution contours. The Chandra data is smoothed M500 (M ) with a Gaussian kernel of 5 pixels where the pixel size is 0.9800. The X-ray emission peaks at the cross (+) location (RA = 10h23m39.9s and d m s Dec = +49 08 38.4 ). The ellipse used for X-ray luminosity measurement Figure 5. The 푃1.4 GHz − 푀500 (bottom) and 푃1.4 GHz − 퐿500 (top) re- is centred at RA = 10h23m40.4s, Dec = +49d08m25.7s with a position an- lations, adapted from Bîrzan et al.(2019). The data points for A990 and gle of 15 degree. The major and minor axes are 4.580 and 6.570 (i.e. 694 kpc RXC J1825.3+3026 are shown by the red diamond and magenta circle, re- and 996 kpc), respectively. spectively. In both panels, the black circles, blue squares, and cyan triangles show radio halos in the samples of Cassano et al.(2013), Martinez Aviles et al.(2016), and Bîrzan et al.(2019), respectively. The blue arrows indicate the power upper limits for undetected halos in the Cassano et al.(2013), Kale et al.(2015), and Bîrzan et al.(2019) cluster samples. The solid lines these variations are still within the mean errors of ∼0.55 keV (see are the best fits for the Cassano et al.(2013) halos. The shadowed regions Figure7, right). This is due to the short exposure duration of the show the 95 percent confidence region. X-ray observations (i.e. 10 ks). Future deep X-ray observations will

MNRAS 000,1–11 (2020) 8 N. D. Hoang et al.

100 100 Relaxed Disturbed Relaxed 10 1 c c w 10 2 1 10 1 10

A990 A990 A990 halo halo halo mini halo mini halo mini halo Disturbed Disturbed no halo 10 3 Relaxed no halo no halo

9 8 7 6 5 9 8 7 6 5 10 3 10 2 10 1 10 10 10 10 10 10 10 10 10 10 w P3/P0 P3/P0

Figure 6. Morphological parameter diagrams for clusters with and without radio halos (Cassano et al. 2010b; Bonafede et al. 2014, 2017). Clusters with radio halos are shown with filled circles. Radio quiet clusters are plotted with cyan squares. Clusters hosting double relics are additionally marked with black circles. −7 The data points for A990 are marked with the red circles. The red dashed lines indicate the median values (i.e. 푐 = 0.2, 푤 = 0.012, 푃3/푃0 = 1.2 × 10 ) −7 calculated from a complete sample of 32 galaxy clusters (Cassano et al. 2010b). The black dotted lines (i.e. 푐 = 0.18, 푤 = 0.021, 푃3/푃0 = 2.5 × 10 ) are calculated for all clusters in the plots. provide information on the possible temperature variations in the In the turbulence re-acceleration model, radio halos are formed cluster. through merger induced turbulence and are likely to be detected in massive merging clusters. In these systems, a fraction of an amount of turbulent energy flux is channelled into the acceleration 4 DISCUSSION of relativistic particles and the amplification/generation of magnetic fields in the ICM during cluster mergers. In less-massive clusters, We have reported the detection of an extended radio source in A990 merging events inject less energy into the ICM, resulting in radio and classify it as a radio halo. The classification is based on (i) its halos being formed less frequently and with steeper spectra (e.g. extended size of 700 kpc that is not associated with any compact Brunetti et al. 2008; Cassano et al. 2010a, 2012). The model predicts sources, (ii) its location at the cluster centre, (iii) its extended emis- that only about 10 − 15 percent of clusters with a mass below 14 sion largely overlaid the X-ray extended emission from the ICM. 6.5 × 10 푀 hosts Mpc-scale radio halos with spectra flatter than 14 A990 with a mass of 푀500 = (4.9 ± 0.3) × 10 푀 (Planck Col- 훼∼−1.5 (e.g. Cuciti et al. 2015). At these masses, giant halos with laboration et al. 2016) is among the less massive clusters known to steeper spectra are expected to be more common and the fraction host a giant radio halo. of these clusters with halos increase up to 40 percent including The power for radio halos is well known to correlate with the steeper spectrum halos (e.g. Cassano et al. 2006, 2010a; Cuciti mass and X-ray luminosity of their host clusters (e.g. Cassano et al. et al. 2015). Therefore, the detection of radio halos in clusters with 14 2013). To examine the correlations in case of A990, we add our data 푀500 . 6.5 × 10 푀 is expected to be more common at low points to the existing 푃1.4 GHz − 푀500 and 푃1.4 GHz − 퐿500 corre- frequencies with LOFAR (Haarlem et al. 2013), MWA (Phase II; lations (see Figure5). Here 퐿500 is the X-ray luminosity measured Wayth et al. 2018), and uGMRT (Gupta et al. 2017). Our discovery within a radius 푅500 at which the total particle density is 500 times of the radio halo in A990 with LOFAR and the fact that it sits below the critical density of the Universe at the cluster redshift. If 훼 = −1.0, the correlations are in line with such expectations. the radio power at 1.4 GHz is 푃 = 1.2 ± 0.1 × 1023 W Hz−1 1.4 GHz The low power of A990 in the 푃 − 푀 and 푃 − which is well below the expected value for a cluster that has a mass 1.4 GHz 500 1.4 GHz 14 퐿500 planes could also indicate a particular phase in the evolution- of (4.9 ± 0.3) × 10 푀 (i.e. 5 times less power than expected) ary state of the cluster. Magnetohydrodynamic (MHD) simulations and X-ray luminosity of (3.66 ± 0.08) × 1044 ergs s−1 (i.e. 4 times of the re-acceleration of cosmic ray electrons by merger induced underluminous in radio power). In case of 훼 = −1.3 (see Sec. 3.1), turbulence by Donnert et al.(2013) modelled the evolution of a the halo power is lower, i.e. 푃 = 6.1 × 1022 W Hz−1 and is 1.4 GHz radio halo power during cluster mergers. The radio halo becomes significantly below the correlations (i.e. 9 and 8 times lower than brighter after the merger and it is fainter in a later merging stage. expected from the 푃 − 푀 and 푃 − 퐿 correla- 1.4 GHz 500 1.4 GHz 500 At the same time the halo spectrum is flattened in the early stage tions, respectively) and the upper limit values for undetected halos of the merger, but is steepened below −1.5 after a few Gyrs of core (Figure5). The difference could be that A990 lies in the poorly- passage. The halo could be observed for a few Gyrs depending on constrained regions for less-massive clusters in the 푃 − 푀 1.4 GHz 500 the observing frequencies which probe different energy of the pop- and 푃 − 퐿 correlations. However, it is still interesting that 1.4 GHz 500 ulation of relativistic electrons and magnetic field strength. At low the halo in A990 is also about 3 times less powerful than the typical frequencies, the halo is observable much longer due to the large life- upper limits for halos that are thought to have ultra-steep spectra time of relativistic electrons with lower energy and weak magnetic (i.e. blue arrows in Figure5). Our morphological analysis of the 14 fields. cluster A990 with a mass of (4.9 ± 0.3) × 10 푀 and its posi- tion in the 푃1.4 GHz − 푀500 and 푃1.4 GHz − 퐿500 diagrams suggest In general, spectral information are thus crucial to understand that A990 can possibly host a radio halo with an ultra-steep spec- the origin and evolution of radio halos. On-going radio surveys trum. However, adequate higher- or lower-frequency observations of a large fraction of the sky such as LoTSS (Shimwell et al. are necessary to constrain its spectrum and confirm the scenario. 2017, 2019) will provide statistical estimates on the fraction of

MNRAS 000,1–11 (2020) LOFAR Detection of a Radio Halo in Abell 990 9

4.00 4.25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 0.3 0.4 0.5 0.6 0.7 0.8 keV keV

T 500 kpc T 500 kpc 49:12:00 49:12:00

10:00 10:00

08:00 08:00 Declination Declination

06:00 06:00

10:24:00 23:48 36 24 10:24:00 23:48 36 24 Right Ascension Right Ascension

Figure 7. Temperature and error maps overlaid with the LOFAR low-resolution (yellow) and Chandra (black) contours. The contour levels are identical to those in Figure4. The cluster has a mean temperature of 5 keV. radio halos in complete samples of clusters down to relatively small coverage of the northern hemisphere, the LoTSS survey will pro- masses. In combination with high-frequency observations with the vide the fraction of clusters of this mass hosting radio halos that ASKAP/EMU (e.g. clusters with declination between 0 and +30◦), will be compared with the theoretical predictions (e.g. Cassano uGMRT and VLA telescopes, the halo spectra will be measured et al. 2010a). A combination with high-frequency observations will which will improve our understanding of the mechanism of particle provide spectral information on the sources that are important for acceleration and magnetic field amplification/generation at cluster studies of the origin of radio halos. scales.

ACKNOWLEDGEMENTS

5 CONCLUSIONS We thank the anonymous referee for the helpful comments. DNH and A.Bonafede acknowledge support from the ERC-Stg17 In this paper we present LOFAR 144 MHz observations, as part of 714245 DRANOEL. MB acknowledges support from the Deutsche the LoTSS surveys, of the galaxy cluster Abell 990 (푧 = 0.144) with Forschungsgemeinschaft under Germany’s Excellence Strategy - 14 a mass of 푀500 = (4.9±0.3)×10 푀 . We detect an extended radio EXC 2121 "Quantum Universe" - 390833306. A.Botteon, EO, and emission with a projected size of ∼700 kpc at the cluster centre. Due RJvW acknowledge support from the VIDI research programme to its location and size, we classify the extended radio emission as with project number 639.042.729, which is financed by the Nether- a radio halo. lands Organisation for Scientific Research (NWO). HR acknowl- The flux density at 144 MHz for the halo is 푆144 MHz = 20.2 ± edges support from the ERC Advanced Investigator programme 2.2 mJy (i.e. above 3휎 contours), corresponding to a radio power NewClusters 321271. AD acknowledges support by the BMBF Ver- 24 −1 of 푃144 MHz = (1.2 ± 0.1) × 10 W Hz (푘-corrected). The radio bundforschung under the grant 05A17STA : The Jülich LOFAR halo is one of the two lowest power radio halos detected to date. Long Term Archive and the German LOFAR network are both Our analysis on the dynamical status of the cluster using the coordinated and operated by the Jülich Supercomputing Centre X-ray data reveals that the cluster has a slightly disturbed ICM and (JSC), and computing resources on the supercomputer JUWELS does not have major merger activities. The X-ray luminosity of the at JSC were provided by the Gauss Centre for supercomputing 44 −1 halo between 0.1–2.4 keV is 퐿500 = (3.66 ± 0.08) × 10 ergs s . e.V. (grant CHTB00) through the John von Neumann Institute for The cluster mean temperature is relatively low, at 5 keV. There Computing (NIC). These data were (partly) processed by the LO- are possible variations in the temperature, but require deep X-ray FAR Two-Metre Sky Survey (LoTSS) team. This team made use of observations to confirm. the LOFAR direction independent calibration pipeline (https:// Our measurements on the radio power, X-ray luminosity, and github.com/lofar-astron/prefactor) which was deployed the ICM temperature are at the low ends. This can be explained by by the LOFAR e-infragroup on the Dutch National Grid infrastruc- the cluster being a dynamically disturbed, less massive system that ture with support of the SURF Co-operative through grants e-infra contains less energy than more massive clusters. 160022 e-infra 160152 (Mechev et al. 2017, PoS(ISGC2017)002). Our study reveals the potential of detecting many faint radio The LoTSS direction dependent calibration and imaging pipeline halos in less-massive galaxy clusters with LOFAR. With the full (http://github.com/mhardcastle/ddf-pipeline) was run

MNRAS 000,1–11 (2020) 10 N. D. Hoang et al. on compute clusters at Leiden Observatory and the University of Freeman P.,Doe S., Siemiginowska A., 2001, Sherpa: a mission-independent Hertfordshire which are supported by a European Research Council data analysis application. pp 76–87, doi:10.1117/12.447161, Advanced Grant [NEWCLUSTERS-321271] and the UK Science http://proceedings.spiedigitallibrary.org/proceeding. and Technology Funding Council [ST/P000096/1]. We have includ- aspx?articleid=892330 ing the lotss-dr2-infrastructure list as this project made use of the Fruscione A., et al., 2006, CIAO: Chandra’s data analysis system. DR2 pipeline and post processing extraction/selfcal scheme. The p. 62701V, doi:10.1117/12.671760, http://proceedings. spiedigitallibrary.org/proceeding.aspx?doi=10.1117/ scientific results reported in this article are based on data obtained 12.671760 from the Chandra Data Archive. This research has made use of soft- Giacintucci S., et al., 2005, A&A, 440, 867 ware provided by the Chandra X-ray Center (CXC) in the application Giovannini G., Tordi M., Feretti L., 1999, New Astron., 4, 141 packages CIAO, ChIPS, and Sherpa. The National Radio Astron- Gupta Y., et al., 2017, Curr. Sci., 113, 707 omy Observatory is a facility of the National Science Foundation Haarlem M. P. V., et al., 2013, A&A, 2, 1 operated under cooperative agreement by Associated Universities, Hardcastle M. J., et al., 2016, MNRAS, 462, 1910 Inc. Hoang D. N., et al., 2019, A&A, 622, 1 Hlavacek-Larrondo J., et al., 2018, MNRAS, 475, 2743 Intema H. T., Jagannathan P., Mooley K. P., Frail D. A., 2017, A&A, 598, DATA AVAILABILITY A78 Kale R., Venturi T., Cassano R., Giacintucci S., Bardelli S., Dallacasa D., The data underlying this article will be shared on reasonable request Zucca E., 2015, A&A, 581, A23 to the corresponding author. Martinez Aviles G., et al., 2016, A&A, 595, A116 Offringa A. R., 2016, A&A, 595, 1 Offringa A. R., van de Gronde J. J., Roerdink J. B. T. M., 2012, A&A, 539, REFERENCES A95 Offringa A. R., et al., 2014, MNRAS, 444, 606 Ackermann M., et al., 2010, ApJ, 717, L71 Perley R. A., Butler B. J., 2017, Astrophys. J. Suppl. Ser., 230, 7 Ackermann M., et al., 2016, ApJ, 819, 149 Petrosian V., 2001, ApJ, 557, 560 Alam S., et al., 2015, Astrophys. Journal, Suppl. Ser., 219 Alden B., Hallman E. J., Rapetti D., Burns J. O., Datta A., 2019, Astron. Pfrommer C., Enßlin T. A., 2004, MNRAS, 352, 76 Comput., 27, 147 Planck Collaboration et al., 2016, A&A, 594, A27 Bacchi M., Feretti L., Giovannini G., Govoni F., 2003, A&A, 400, 465 Poole G. B., Fardal M. A., Babul A., McCarthy I. G., Quinn T., Wadsley J., Balucinska-Church M., McCammon D., 1992, ApJ, 400, 699 2006, MNRAS, 373, 881 Ben Bekhti N., et al., 2016, A&A, 594, A116 Rines K., Geller M. J., Diaferio A., Kurtz M. J., 2013, ApJ, 767, 15 Bernardi G., et al., 2016, MNRAS, 456, 1259 Rudnick L., Lemmerman J. A., 2009, ApJ, 697, 1341 Bîrzan L., et al., 2019, MNRAS, 487, 4775 Santos J. S., Rosati P., Tozzi P., Böhringer H., Ettori S., Bignamini A., 2008, Böhringer H., et al., 2010, A&A, 514, A32 A&A, 483, 35 Bonafede A., et al., 2014, Mon. Not. R. Astron. Soc. Lett., 444, L44 Savini F., et al., 2019, A&A, 622 Bonafede A., et al., 2015a, MNRAS, 454, 3391 Scaife A. M. M., Heald G. H., 2012, Mon. Not. R. Astron. Soc. Lett., 423, Bonafede A., et al., 2015b, MNRAS, 454, 3391 30 Bonafede A., et al., 2017, MNRAS, 470, 3465 Shimwell T. W., et al., 2017, A&A, 598, A104 Bonnassieux E., Tasse C., Smirnov O., Zarka P., 2018, A&A, 615, A66 Botteon A., et al., 2019, A&A, 630, A77 Shimwell T. W., et al., 2019, A&A, 622, A1 Brunetti G., Jones T. W., 2014, Int. J. Mod. Phys. D, 23, 1430007 Smirnov O. M., Tasse C., 2015, MNRAS, 449, 2668 Brunetti G., Lazarian A., 2007, MNRAS, 378, 245 Smith R. K., Brickhouse N. S., Liedahl D. A., Raymond J. C., 2001, ApJ, Brunetti G., Setti G., Feretti L., Giovannini G., 2001, MNRAS, 320, 365 556, L91 Brunetti G., et al., 2008, Nature, 455, 944 Tasse C., 2014a, ArXiv e-prints 1410.8706v1 Brunetti G., Zimmer S., Zandanel F., 2017, MNRAS, 472, 1506 Tasse C., 2014b, A&A, 566 Buote D. A., Tsai J. C., 1995, ApJ, 452, 522 Tasse C., et al., 2018, A&A, 611, 1 Cassano R., Brunetti G., Setti G., 2006, MNRAS, 369, 1577 Venturi T., Bardelli S., Morganti R., Hunstead R. W., 2000, MNRAS, 314, Cassano R., Brunetti G., Rottgering H. J. A., Bruggen M., 2010a, A&A, 594 509, 68 Venturi T., Bardelli S., Dallacasa D., Brunetti G., Giacintucci S., Hunstead Cassano R., Ettori S., Giacintucci S., Brunetti G., Markevitch M., Venturi R. W., Morganti R., 2003, A&A, 402, 913 T., Gitti M., 2010b, ApJ, 721, L82 Wayth R. B., et al., 2018, Publ. Astron. Soc. Aust., 35, e033 Cassano R., Brunetti G., Norris R., Röttgering H., Johnston-Hollitt M., Trasatti M., 2012, A&A, 548, A100 Wen Z. L., Han J. L., 2013, MNRAS, 436, 275 Cassano R., et al., 2013, ApJ, 777, 141 Werner N., Urban O., Simionescu A., Allen S. W., 2013, Nature, 502, 656 Cassano R., et al., 2019, ApJ, 881, L18 Williams W. L., et al., 2016, MNRAS, 460, 2385 Condon J. J., Cotton W. D., Greisen E. W., Yin Q. F., Perley R. A., Taylor Zandanel F., Pfrommer C., Prada F., 2014, MNRAS, 438, 124 G. B., Broderick J. J., 1998, Astron. J., 115, 1693 Zhu Z., et al., 2016, ApJ, 816, 54 Cuciti V., Cassano R., Brunetti G., Dallacasa D., Kale R., Ettori S., Venturi de Gasperin F., et al., 2019, A&A, 622, A5 T., 2015, A&A, 580, A97 van Weeren R. J., et al., 2016, Astrophys. J. Suppl. Ser., 223, 2 Donnert J., Dolag K., Brunetti G., Cassano R., 2013, MNRAS, 429, 3564 van Weeren R. J., de Gasperin F., Akamatsu H., Brüggen M., Feretti L., Duchesne S. W., Johnston-Hollitt M., Offringa A. R., Pratt G. W., Zheng Q., Kang H., Stroe A., Zandanel F., 2019, Space Sci. Rev., 215, 16 Dehghan S., 2017, 35, 1 van Weeren, R. J., Shimwell, T. W., Botteon, A., et al. 2020, ArXiv e-prints, Feretti L., Giovannini G., Govoni F., Murgia M., 2012, Astron. Astrophys. arXiv: 2011.02387 Rev., 20, 54 Frank K. A., Peterson J. R., Andersson K., Fabian A. C., Sanders J. S., 2013, ApJ, 764, 46

MNRAS 000,1–11 (2020) LOFAR Detection of a Radio Halo in Abell 990 11

APPENDIX A: OPTICAL COUNTERPARTS OF COMPACT SOURCES To identify optical counterparts of the radio compact sources, we overlaid the LOFAR high-resolution contours on SDSS images in Figure A1.

This paper has been typeset from a TEX/LATEX file prepared by the author.

MNRAS 000,1–11 (2020) 12 N. D. Hoang et al.

Source A Source B 49:07:30 Source C 49:09:00 49:08:15 00

08:45 00 06:30 Declination Declination Declination 30 07:45 00

15 30

10:23:45 44 43 42 41 40 39 10:23:43 42 41 40 39 38 10:23:42 39 36 33 Right Ascension Right Ascension Right Ascension Source D Source E Source F 49:07:15 49:07:30 49:10:00

00 00 09:30 Declination Declination

Declination 06:45 06:30 00

30

00 08:30

10:23:50 49 48 47 46 45 44 10:24:09 06 03 00 10:24:00 23:57 54 51 Right Ascension Right Ascension Right Ascension

Source G and H 49:12:00 Source I 49:10:45 Source J 49:12:30

11:40 30 00

20 15 Declination 11:30 Declination Declination

00

00 00 10:40

10:23:51 48 45 42 10:23:30 28 26 24 10:23:25 24 23 22 21 20 Right Ascension Right Ascension Right Ascension

49:06:00 49:10:15 Source K Source L Source M 49:04:40

05:45 00

20

30 09:45 Declination Declination Declination 00

15 30

03:40

10:23:26 25 24 23 22 21 10:23:20 19 18 17 16 15 10:23:36 34 32 30 Right Ascension Right Ascension Right Ascension

Source N Source O 49:05:15 Source P 49:05:30 49:06:45

00 15 30

04:45 Declination 00 Declination Declination 15

30 04:45 00

10:23:46 45 44 43 42 41 10:23:55 54 53 52 51 50 10:24:01 00 00 23:58 57 56 Right Ascension Right Ascension Right Ascension

Figure A1. SDSS optical (i, r, and g band) images overlaid with the LOFAR high-resolution contours. The contours are at the same levels as those in Figure1. The source labels are given in Figure2.

MNRAS 000,1–11 (2020)